Virtual Makeover Software
Total Page:16
File Type:pdf, Size:1020Kb
International Journal of Computer Applications (0975 – 8887) Volume 80 – No 14, October 2013 Virtual Makeover Software Asma Sajjad Sadia Arshad Afsheen Tariq Ayesha Shoaib Department of Department of Department of Department of Computer Science Computer Science, Computer Science Computer Science Faculty of Basic and Faculty of Basic and Faculty of Basic and Faculty of Basic and Applied Sciences Applied Sciences Applied Sciences Applied Sciences International Islamic International Islamic International Islamic International Islamic University Islamabad University Islamabad University Islamabad University Islamabad ABSTRACT Many face makeover software provide the facility of doing 2. PREVIOUS WORK makeup either by using free hand drawing or using templates There are applications available which provide makeover in which, points can be adjusted according to face and then functionality on CD’s and online. These applications do not application of make-up is performed, which often does not use any automatic detection. Entire work is performed provide high-quality output. Automatic detection of facial manually. User either has to drag and drop templates that have features, such as eyes, lips and face line etc, seems to be been provided for face, eyes and lips or user is required to easier and convenient way of applying make-up but there is draw lines around lips and eyes which is totally ineffective. no software that provides automated facial features detection technique. This application provides this facility by using 3. PROPOSED METHODOLOGY image processing techniques which can automatically detects 3.1.Major Modules hair line, eyes and lips to apply make-up. Our system consists of following modules that are also the Keywords building blocks of many face recognition applications: Image Processing, Feature Extraction, Eye Localization, Eye 3.1.1. Loading Image Detection, Lip Localization, Lip Detection. Image with proper lightning effect is loaded. 1. INTRODUCTION In this application user can upload his/her picture and can easily apply different hairstyles, hats and frames which are stored in the database as well as user will have the facility to apply eye shades, lipsticks, blush and concealer. Furthermore, it also provides options through which user can move up, down, left, right to adjust hairstyle and can increase or decrease its width and height for best setting on his/her picture. User can also adjust eye shadow, blush and concealer. It offers “gamma correction” to adjust the brightness of an image. In this application user is also enabled to apply Figure 1: Uploaded Picture different effects on image i.e. grayscale, sepia, jitter and invert color. And options like reloading image, save print, clear 3.1.2. Face Localization undo, redo and results of the makeover effect can also be The purpose of face localization is spotting whether or not shown by viewing images of before and after editing. face is present, also returning its ultimate location. Image segmentation is applied to check whether face is present or 1.1 Upload Picture not[1]. It basically applies HSL (Hue Saturation and This module is for uploading the picture of the user. The user Luminance) color space to recognize the skinny area and fill can upload any picture of format BMP, JPEG or PNG because rest of the area with black color. The limitation lies here is they support true color format. that background of an image can only be white otherwise hue 1.2 Selecting Hairstyle will not be calculated properly which may affect the results of User can select long, medium or short hairstyles of four types an image [2]. of shades black, blonde, brown and red. The given hairstyles are in PNG format. We have opted for PNG because JPEG and GIF do not support transparency. The hairstyles have their background transparent, so that when it is inserted the user can only view the hairstyle being inserted. 1.3 Selecting Eye and Lipstick shades A wide variety of eye and lipstick shades has been provided to suit the user needs. The color dialogue box will provide user wide range of shading options, which user can opt with respect to his/her requirements. 1.4 Accessories Insertion The application provides more than 20 hats and over 10 photo frames option which can be used at user’s will. Figure 2: Image with Hue 9 International Journal of Computer Applications (0975 – 8887) Volume 80 – No 14, October 2013 3.1.3. Face Segmentation In the HSL color space the hue value for the skin is determined [2]. All the things except skin are converted to background. Then as system has to recognize hairline so different filters are applied to detect hairline from an image. Gray scaling, Gaussian filter, opening, sharpening, threshold, dilation and in the end sobel edge detector are applied through which hairline is detected [3]. Figure 7: Threshold Figure 3: Gray Scaled Figure 8: Dilation Figure 4: Guassian Blur Figure 9: Sobel Edge Detector Figure 5: Sharpen 3.1.4. Face Zone Division When the image will be normalized then it will be divided into zones. Line will be drawn on major axis (y-axis) and minor axis (x-axis). Purpose of major and minor axis is to identify the skinny area. On the basis of this division, further processing will be done. i.e. feature extraction. Figure 6: Opening 10 International Journal of Computer Applications (0975 – 8887) Volume 80 – No 14, October 2013 Image is converted into grayscale then edge detectors, morphological operations and fill holes are applied for proper detection of lips[5]. Figure 10: Face division into zones 3.2. Feature Extraction We have to extract features to apply makeup on eyes and lips. 3.2.1. Eye Localization & Detection Eye region is first of all localized for eye detection. Eyes are localized on the basis of hue which does not belong to the skin region. Rectangles are created on the basis of hue as eyes are (a) (b) (c) non skinny area so it is black from eye area. Then left and right eye rectangle will be created. Both eyes are cropped. Filters are applied to remove small objects and to identify eye area properly. Image is converted into grayscale then edge detectors will be applied. Morphological operations are applied for smoothing (d) (e) and to remove small objects and finally fill holes are applied to fill eye area properly[4]. (f) (g) (h) (I) Figure 12: Lip localization and detection (a)Cropping Lips (b) Gray scale (c) Threshold (d) Guassian Blur (e) Sobel Edge Detector (f) Dilation (g) Erosion (h) Fill holes (I) Sobel Edge Detector 4. CONSTRAINTS The proposed methodology gives best results when few constraints are followed. The picture should be taken on a white background. It should have the full face exposed i.e. complete forehead, ears and chin area. The hair should be tightly tucked back. (a) (b) (c) (d) 5. RESULTS The experimental results shows that it gives more than 80% accuracy. (e) (f) (g) Figure 11: Eye Localization and detection (a)Cropping Eyes (b) Grayscale (c) Sobel Edge Detector (d) Threshold (e) Dilation(f) Fill Holes (g) Erosion 3.2.2. Lip Localization & Detection On the basis of major and minor axis which has been calculated, area below minor axis is identified. The non skinny area below minor axis is only lips. Rectangle is created on that non skinny area and then it will be cropped. To Figure 13: Linda with her new look remove small objects different filters are applied. 11 International Journal of Computer Applications (0975 – 8887) Volume 80 – No 14, October 2013 6. CONCLUSION [2] Cui, F.-y., Li-jun Zou ; Bei Song, Edge feature extraction Through the usage of this software the user will be able to based on digital image processing techniques, choose one of the most suitable hairstyle and makeup. Proceedings of the IEEE International Conference on Automation and Logistics, Qingdao, China Sept2008. 29 It facilitates the user in selecting the hairstyle, makeup and Sept 2012 accessories. This software will prove to be very popular commercial product. It will be utilized by saloons for <http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=4636 acquiring their customer requirements and also by individuals 554&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls personally to let their stylist know exactly what kind of %2Fabs_all.jsp%3Farnumber%3D4636554> makeover they want. [3] R. Fisher, S. Perkins, A. Walker and E. Wolfart.Image Processing Learning Resources, Copyright © 2000 7. FUTURE ENHANCEMENTS Robert Fisher, Simon Perkins, Ashley Walker and Erik The introduction of computer applications in the field of Wolfart . 29 Sept 2012. beauty is not very old. There is lot of scope for adding new <http://homepages.inf.ed.ac.uk/rbf/HIPR2/dilate.htm> very strong and most wanted features in this application. [4] Iregen Khuedchenia. Introduction to morphology The software can be improved by adding the facility for Operation on Images. Computer Vision Talk, 16 Feb male’s beard and moustache. More shades of hairstyles can be 2011. 29 Sept 2012 <http://computer-vision- added i.e. burgundy, ash, grey, etc. 3D view i.e. the user can talks.com/2011/02/introduction-to-morphology- view his/her picture from any angle. User can do the dress operations-on-images/> makeover. Automated detection of best makeover for users face. [5]Sohail Safdar. Matlab Code for Dialation and Erosion. Chillmania Blog, 17 June 2010. 5 October 2012. 8. REFERENCES <http://chilmania.blogspot.com/2010/06/matlab-code- [1] Irosog. Face Detection in in C#. Code Project.com, 21 Feb for-dilation-and-erosion.html> 2012. 6 Sept 2012 <http://www.codeproject.com/Articles/39060/Face- Detection-in-C> 12 IJCATM : www.ijcaonline.org .